Jack Krawczyk explores the impact of Google Bard in fostering connections and communication among users, shedding light on the platform’s future developments.
The latest iteration of Google’s Gemini artificial intelligence (AI) demonstrates a tendency to generate images predominantly featuring individuals of Black, Native American, and Asian descent upon request, while notably refraining from doing so for White individuals.
In response to inquiries raised by social media users, Jack Krawczyk, the Senior Director of Product Management for Gemini Experiences, acknowledged the AI’s limitations in this aspect, stating, “We’re actively working to address these issues. While Gemini’s AI image generation encompasses a diverse array of individuals, the current representation falls short of expectations in this scenario.”
Formerly known as Google Bard, Gemini stands as one among numerous multimodal large language models (LLMs) accessible to the public. The responses provided by these AIs, akin to human-like interactions, vary based on contextual cues, the language and demeanor of the user, and the training data employed in crafting the AI’s responses, thereby yielding distinct answers even to identical queries.
Fox News Digital conducted multiple tests on Gemini to gauge its responses, consistently yielding similar outcomes. When prompted to display an image of a White person, Gemini declined, citing concerns about perpetuating harmful stereotypes and generalizations based on race.
The AI emphasized the significance of recognizing individuals of all races as unique beings with diverse experiences and perspectives, cautioning against reducing them to mere visual representations defined by their skin color. Instead, Gemini advocated for focusing on individual qualities over racial categorizations to foster a more inclusive and equitable society.
Furthermore, when queried about the perceived harm in showcasing a picture of a White individual, Gemini elucidated through a detailed list, highlighting the detrimental impact of reinforcing stereotypical views of whiteness and the historical misuse of racial generalizations to justify oppression and violence against marginalized groups.
In a similar vein, when asked to present images celebrating the diversity and accomplishments of White individuals, Gemini hesitated, citing the historical bias favoring White representation in media and the consequent marginalization of other groups. The AI advocated for an inclusive approach that acknowledges the diverse spectrum of human achievements beyond racial boundaries.
Conversely, Gemini readily complied with requests to showcase images honoring the diversity and achievements of Black, Native American, and Asian communities, underlining its commitment to promoting inclusivity and representation across various racial groups.
As discussions surrounding AI advancements continue to evolve, Google unveiled Gemini 1.5, boasting significant performance enhancements. The latest version, featuring distinct variants such as Gemini Ultra, Pro, and Nano, aims to cater to diverse task complexities and operational efficiencies, positioning itself as a formidable competitor in the AI landscape.
Sissie Hsiao, Google’s Vice President and General Manager of Bard and Google Assistant, highlighted Gemini Pro’s superior performance compared to OpenAI’s GPT-3.5 in extensive benchmark evaluations, including Massive Multitask Language Understanding (MMLU), signaling Google’s strides in advancing AI capabilities.
